CONTRIBUTED BY: THOMAS HIGHAM

When the C14 method was originally developed, Libby and his research team had to assume that the ratios of the carbon isotopes they were measuring had been altered only by 14C decay (Taylor, 1987:3) and that the sample material accurately represented the event to be dated. Sample materials deposited in archaeological or geological contexts seldom remain in pristine condition, of course, they are often degraded and altered chemically. Libby realized that the residual carbon 14 of some samples would be thus affected and suggested that some materials would be more accurate for dating than others. He predicted that charcoal would be the most effective, shell, the least.
The following types of sample have been commonly radiocarbon dated:

Since the 1950's, a number of researchers have concentrated on investigating and reducing the effects of this post-depositional contamination. This field of inquiry is known as sample pretreatment and it is concerned with removing post-depositional contaminants by isolating sample fractions containing carbon which is autochthonous and therefore accurately dates the event in question.

This section is divided into a number of areas:

Introduction

Submission of samples:

 The key issue in sample pretreatment is that there is no method, or methods, that can be universally applied to all types of material from archaeological or geological contexts. Pretreatments are designed to remove the contaminating substances that have affected the sample during its post-depositional history. If pretreatments were able to be uniformly implemented, there would have to be a uniform and predictable array of post-depositional characteristics between all samples. Clearly, this is not the case. Each sample submitted for dating has its own specific depositional history. The variety in environment and post-depositional features is reflected in the variety and complexity of pretreatment procedures and the variety of different types of dateable material (see above). Nevertheless, there are certain laboratory procedures which are associated with specific sample types and environments, and a number of accepted and often repeated pretreatment methods. These are described in detail below.

The laboratory decides on the most effective pretreatment procedure through a careful examination of each submitted sample. A number of variables feature in this consideration, one of the most important concerns the environment within which the sample was deposited. The lab must consider the possibility of contamination in each sample it dates and depends upon information supplied by the submitter and collector of the material for its assessment. The submitter should supply information detailing the type of environment from which the sample was obtained and commenting on the presence of rootlet intrusion and contaminants.

There is more information about AMS Sample Preparation and pretreatment at the Oxford AMS lab and the Rafter Radiocarbon Lab.

The submitter should also describe the relationship between the material and the geological, or archaeological context to be dated. A stratigraphical diagram should be drawn to enable the dater to understand completely the site and origin of the material, and to consider the ability of the lab to adequately date the sample in question. The submitter should also indicate the degree of accuracy and precision required. Sometimes, a precise date is not needed and pretreatment methods designed to reduce errors will not be necessary. Many commercial laboratories have different charges for dating depending on the precision (and speed) that is required. A high-precision date may involve the lab in more intensive pretreatment and labour and consequently costs are higher.

Submitters should send as much sample as possible because of the destructive nature of certain pretreatment techniques. This is particularly relevant for laboratories which use conventional methods of dating. Bone dating, for example, requires large amounts of sample because the fractions which are usually extracted comprise a small percentage of the total material and the target fractions decompose rapidly. Often, submitted samples are divided and one portion retained as a reference in case the original sample is lost, or a further date required.

Sometimes, submitters perform basic pretreatments, usually involving a wash in distilled water and the removal of root material. This should be reported in the submission forms accompanying samples sent to the laboratory.

Contamination

Contamination may be artificially or naturally caused. Artificial contamination may be blamed on human negligence during the collection and processing of samples. Contaminants often include ash from tobacco, hair and fibres, paper from packing material and oil or grease (Hogg, 1982:21). Natural contamination occurs in the post-depositional environment. Samples may be contaminated by material which make any radiocarbon result either too old or too young. The most common source of contamination by modern carbon is caused by rootlet intrusion. Organic samples such as wood, charcoal, soil and bone are especially prone to this and should be examined closely before, and after collection, for evidence of root penetration (Hogg, 1982:18). Contamination may also be caused by humic acids circulating throughout the soil. Humics are the decayed remnants of dead plants. They may exchange carbon or adhere to samples that have large surface areas and make any radiocarbon results too young. This surface exchange is termed 'adsorption' and is especially common in samples such as peats, charcoals and muds.

Certain samples, especially shell, may show evidence of isotopic exchange or recrystallisation. Isotopic exchange occurs when shells exchange carbon with percolating ground acids. This alters their isotopic ratios and affects their 'true' age. The exchange usually occurs on the exterior shell surfaces in terrestrial environments and is common in samples found below the water table (Hogg, 1982:18). An analysis of the carbon isotope ratios using a mass spectrometer will reveal the extent of any exchange. A correction factor may be applied or the sample rejected on this basis. Recrystallisation refers to the modification of shell aragonite to calcite, often involving an exchange with modern calcite and a subsequent altering of isotopic ratios.

In investigating the extent and effect of these types of sample contamination, the radiocarbon laboratory is faced with two major problems. First, it has to identify precisely the nature and size of contamination. Second, it needs to assess its magnitude and direction of change (Gupta and Polach, 1985:129).

Olsson (1974), and Gupta and Polach (1985:129-134) have considered the nature of this relationship between sample, contaminant and magnitude of error. They suggested that by "guesstimating" the age difference between the 'true' sample age and that of the contaminant, and calculating the relative size of the contaminant in the sample, it was possible to determine the extent of the error caused by the contaminant and apply a correction. A set of graphs were shown by the authors' to show the range of errors associated with samples contaminated by older and younger carbon. In general, the older the sample, the greater became the effect of contamination, even if the percentage of contaminant to sample was small (see Table 1). This consideration is based on the assumption that any contamination was instantaneous (Gupta and Polach, 1985:130). More complex problems arise when the contaminating fraction changes temporally in size.

% Contamination by modern carbon
0% 1% 5% 10% 25% 50%
At 900 BP 900 (0) 890 (1) 850 (5) 810 (11) 670 (26) 440 (51)
% Contamination by old carbon
At 900 BP 0% 1% 5% 10% 25% 50%
900 (0) 980 (9) 1320 (47) 1770 (97) 3280 (264) 6630 (637)


Table 1: The effect of contamination by old and modern carbon upon a sample with a 'true age' of 900 BP. Figures in brackets give the % error introduced by the contaminant (table from Caughley, 1988).

The other major issue in sample contamination concerns samples which contain small errors. Dates which are clearly too old or young are easily recognised and investigated, those that contain less significant errors are more difficult to identify. According to Olsson (1979), the danger is that these dates will often be considered to be reliable when they are not because they fall close to the expected age.

Assessing the effect of treatment

There are a number of ways to guage the effects of pretreatment upon samples. The most common is to analyse and date the different fractions that have been removed. This will reveal the magnitude of error. If the dates from sample and contaminant are close, the associated error will be insignificant, however, if there is a major difference, the errors will rise. If different fractions give statistically identical results then it is concluded that no significant contamination has occurred. Another more complex method, concerns the analysis of the chemical properties of successively removed fractions. This enables the investigator to identify the types of contaminants present and their concentration, and determine the success of pretreatments. This quantitative method targets specific contaminants and measures them as a percentage of the total sample. It is a costly measure and is usually confined only to research programmes or to very important samples. A third method is cross-checking. The original date is 'checked' by dating other contemporaneous materials. Similarly, the date may be assessed using reliable cultural and stratigraphic markers as chronological reference points. For example, in Mycenean, Minoan and Cycladic archaeology, reliable chronological markers exist in the form of pottery, which varies both stylistically and temporally through the bronze age, providing a useful reference. A final, though less reliable method, is the analysis of the amount of carbon dioxide produced in combustion or hydrolysis, in proportion to the size of the sample. The rationale here is that successful pretreatments should maximise the amounts of pure carbon dioxide produced, proportional to sample size. The assumption is that the CO2 produced is not contaminated.

By dating the fractions removed, then, it may be possible to consider whether pretreatments have revealed a 'truer' age. An example of such a study may be seen in the recent work of Head et al. (1989), who implemented a geochronological research programme in loess deposits in China, near Xian. Humic contamination was expected because of the discovery of rootlet intrusion and the muddy and wet texture of the deposit. Humic acids are mobile, decay products from recently dead plants which leach down through site profiles, being adsorbed by certain receptive substances and affecting 'true' dates. The usual method for removing humic acids is through treatment with base solutions, most commonly, sodium hyroxide (NaOH). Often, by increasing the polarity of these washes, more effective removal of humics is possible. Each wash results in a base soluble fraction and a base insoluble fraction. Both can be analysed for contaminants, or dated. If dates from the two fractions vary significantly then it may be assumed that some degree of contamination has occurred. Sometimes the base insoluble fraction will be deemed the most reliable fraction, because the base pretreatments will have extracted the humic acid contamination. Conversely, the base soluble fraction may be considered more reliable, because the humic component will be a terrestrial humic acid, that is, a contemporaneous fraction of the peat. Deciding which sample fraction was reliable in Xian was made by analysing a number of complex variables, for example; the elemental atomic ratios for the fractions, which indicate the types of humic substances in the stratum; optical density measurements, which reveal approximately the time a humic substance has been present in the sample; solvent extraction analysis, which may show the type of contaminant residue; and 14C determinations, which enable comparison of the dates from the two fractions. Through these analyses it was possible for Head et al to guage the effect and origin of contamination and estimate the success of the pretreatment. They found in this case, that the base soluble fraction was the most reliable for dating. The insoluble fraction revealed ages that were considered too young. The clay component which formed the basis of the insoluble fraction was providing a matrix suitable for adsorption of contaminants and attracting young humic substances (Head et al, 1989; 685-689).

There are two major types of pretreatment applicable to carbonaceous samples:

INDEX | Introduction | Measurement | Applications | WWW Links | k12 | Publication | Corrections |
Age calculation | Calibration | Pretreatment | References | Awards | Credits | What's new? | Email |

HTML DOCUMENT BY T.HIGHAM.